Storage Policies on SeaWulf

Storage Policies

SeaWulf provides several storage areas for different types of data and workflows. The table below outlines their intended use, retention policies, and capacity limits.

Storage Areas

Path Storage Quota File Quota Backed Up? Retention Policy Description
/gpfs/home/<netid> 20 GB ~200K files Yes Persistent Personal directory for scripts, configs, and small files.
/gpfs/scratch/<netid> 20 TB 10 million files No 30-day purge High-capacity temporary space for large datasets. Files older than 30 days are deleted.
Attempts to bypass limits may result in loss of cluster access.
/gpfs/projects/<groupname> Custom (typically up to 10 TB) ~10 million files No Until project end Shared space for research groups. Use for workflows and shared datasets. Not for long-term storage.
/gpfs/software/ System-managed N/A N/A Permanent System-wide applications and libraries. Access via module load.

Project Spaces

PIs can request project space to share data with their research group. For example, the "Smith Project" space would be located at /gpfs/projects/SmithGroup.

Use project space for datasets, workflows, and results that the whole group needs. It's not meant for long-term archival storage. Move any data that must be retained indefinitely to another system. Since project space isn't backed up, PIs are responsible for maintaining their own backups.

PIs can request up to 10 TB during the initial project setup or later through the ticketing system. Include the following details:

  • Requested amount of space
  • Description of the data and usage (access patterns, bandwidth needs)
  • Estimated duration of need (not for permanent storage)
  • Confirmation that the space is not backed up and backups are the group’s responsibility

Requests exceeding 10 TB require detailed justification and may only be approved for a limited time.

For information on checking usage and quotas, see Checking Your Storage Quota (myquota).